Prevention of invasive Candida infections in preterm infants: the time is now

Figure 1. Neurodevelopmental impairment and bloodstream infection in infants weighing under 1000 g.

Neurodevelopmental impairment: one or more of the following: PDI or MDI of under 70, cerebral palsy, visual or hearing impairment. Clinical infection: late-onset infection with negative cultures following antibiotic treatment for at least 5 days.

^*p ≤ 0.001 compared with no-infection group.

CoNS: Coagulase-negative staphylococci; MDI: Mental developmental index; PDI: Psychomotor developmental index.

Figure 2. Invasive Candida infection surveillance chart.

^*Other infections are peritoneal and/or other sterile body fluid infections.

ICI: Invasive Candida infections (bloodstream, UTI, meningitis and other infections); NDI: Neurodevelopmental impairment (one or more of the following: psychomotor developmental index or mental developmental index < 70, cerebral palsy, blindness or need for hearing aids/deafness); UTI: Urinary tract infection.

Scope of the problem

There are nearly 30,000 preterm infants weighing under 1000 g and/or 27 or fewer weeks gestation born each year in the USA (National Vital Statistics, US CDC, 2004), translating into approximately 2000–3000 invasive Candida infections (ICIs), 200–400 Candida-related deaths and 900–1200 infants developing neurodevelopmental impairment (NDI) associated with ICI [1,2].

Prevention of ICI is critical since neurodevelopmental impairment or death occurs in 73% of infants weighing under 1000 g [3,4]. In a multicenter analysis by the National Institute of Child Health and Human Development (NICHD) of infants weighing under 1000 g, the mortality rate was 26% and NDI occurred in 57% of survivors with Candida bloodstream infections and 53% with meningitis infections [4]. Compared with other infections, Candida bloodstream infections (BSIs) have the highest associated NDI (Figure 1)[3].

In infants weighing under 1000 g with ICI, mortality rates range from 23 to 66% of the control patients in prophylaxis studies [5–15]. There is a marked difference in mortality between infants weighing under 1000 g and larger infants with ICI. In infants weighing under 1000 g with ICI, a recent analysis using ICD-9 codes reported a crude mortality rate of 26% compared with 13% without candidiasis; for infants weighing over 1000 g with ICI, mortality was 2% compared with 0.4% among infants without candidiasis [1,16,17].

Several studies have advocated empiric antifungal therapy in high-risk patients, with an improvement in mortality in infants weighing under 1500 g [18,19]; although, for infants weighing under 1000 g in the aforementioned NICHD study, prompt or empiric therapy did not decrease NDI or mortality in these patients [4].

In addition to the morbidity and mortality studies, two recent case–control studies have examined the effect of ICIs on cost of hospitalization and length of hospital stay [16,17]. These studies are limited in that they are based on ICD-9 codes, which may not code for all ICIs. The mean increase in hospital costs was US$39,045 for infants weighing under 1000 g, with no difference in length of stay and, for infants weighing at least 1000 g, there was an increase of US$122,302, with an additional length of stay of 16 days [17].

Incidence of ICIs

Several studies have examined ICIs in neonates, reporting a varying incidence between centers and a change of incidence over time. In the largest analysis, data from the National Nosocomial Surveillance System hospitals from 1995 to 2004 (including 128 neonatal intensive care units [NICUs] and 130,523 neonates) indicated that, for infants weighing under 1000 g at birth, 50% of NICUs had fungal BSI rates of at least 7.5%, and that 25% of NICUs had rates over 13.5% [1]. However, there was considerable variation between NICUs, ranging from 3 to 23% for the 10th and 90th percentiles. During a shorter time period (1998–2001), a smaller multicenter study reported similar results when examining BSIs and meningitis [2]. Rates may differ, largely due to the demographics of admitted patients, resuscitation practices, the surgical population and feeding and antibiotic practices; unfortunately, one or more of these factors are lacking from all epidemiologic studies to date. This is exemplified in two recent studies in the UK; while overall rates of ICI in infants weighing under 1000 g were reported to be 2.1% in the UK, antifungal prophylaxis was not ascertained for that study but, in a follow up survey, almost 30% of NICUs used antifungal prophylaxis [20,21]. These were also the larger NICUs and possibly represent a higher percentage of the UK’s extremely preterm infants. Therefore, the rates in the UK may be significantly lower than have been found in the USA and other studies due to the use of antifungal prophylaxis.

What is the incidence of ICI in your NICU?

Most studies only report BSIs and fail to account for meningitis, urinary tract infections (UTIs) (of which a third have renal abscess involvement) and sterile body fluids, such as peritoneal infections, which may occur with necrotizing enterocolitis (NEC) and focal bowel perforations.

Weitkamp et al. recently reported the importance of examining ICI incidence in each NICU, as many flaws exist when solely relying on the literature for incidence rates [15]. The NICU reported that their rate of Candida BSI was 6.8%, while the rate of all ICIs was 10% for infants weighing under 1000 g.

Gestational age has a more linear relationship with ICI prevalence compared with birth weight, and aids in defining the highest-risk patients [20,22–24]. For example, growth charts demonstrate that a 24-week gestation infant could weigh between 468 and 940 g (3rd to 97th percentiles) [25]. Examining the incidence of ICIs by each gestational age as well as by birth weight (by each 100 g), NICUs should be able to see where their rates fall to zero. Figure 2 illustrates an infection-control approach for each NICU to analyze their incidence of ICIs with infection-related mortality and NDI.

Prophylaxis

Fluconazole

Fluconazole prophylaxis targets all sites of potential colonization and dissemination (skin, GI and respiratory tracts and intravenous catheters [central or peripheral] when given via intravenous administration [via a central venous catheter, if present]) [5]. Fluconazole prophylaxis may achieve even greater safety and efficacy in neonates compared with immunocompromised pediatric and adult patients, due to the ability to start prophylaxis shortly after birth (prior to colonization and proliferation) and the need for a much shorter duration of prophylaxis (4–6 weeks in neonates compared with 75–100 days in bone marrow transplant patients).

The efficacy and safety of fluconazole prophylaxis in preterm infants has been reported in over 2600 patients from four randomized controlled trials (RCTs) [5–7,10] and seven retrospective studies [8,9,11–15], with no significant adverse effects or emergence of resistance reported.

The first RCT targeted high-risk infants weighing under 1000 g with a central venous catheter or endotracheal tube in place during the first 5 days of life and continued prophylaxis while they still needed intravenous access (peripheral or central) for up to 6 weeks [12]. ICI occurred in none of the fluconazole-treated infants compared with 20% of the placebo-treated patients (p = 0.008). Subanalysis demonstrated efficacy even in the most extremely preterm infants. In total, 9% of study infants were aged less than 24 weeks gestation and fluconazole prophylaxis was effective in preventing ICI in this group (p = 0.04). Furthermore, even when these extremely high-risk patients were removed from the analysis, fluconazole prophylaxis was still effective in decreasing the incidence of ICI (p = 0.01). Recently, a multicenter RCT in 322 neonates weighing under 1500 g demonstrated a decrease in ICIs in fluconazole-treated infants (odds ratio [OR]: 0.25; 95% confidence interval [CI]: 0.10–0.59; p = 0.001) [10]. In a secondary analysis, prophylaxis was efficacious for infants less than 27 weeks gestation (p = 0.007) and weighing under 1000 g (p = 0.02).

An important feature of these RCTs was the primary use of a different antifungal agent (amphotericin B), rather than fluconazole, for the treatment of documented fungal infections (as well as suspected or empiric treatment) for any infant in the NICU. This reduces the overall exposure of fungi to fluconazole and reduces the potential for resistance to develop [26]. It also assures appropriate antifungal therapy if an infant on prophylaxis develops an infection due to a resistant organism.

Meta-analysis using Mantel–Haenszel methods of the studies from 2001–2007 demonstrated that fluconazole prophylaxis reduced the risk of developing ICIs in high-risk infants weighing under 1000 g by 91% (OR: 0.09; 95% CI: 0.04–0.24; p = 0.0004) and in all infants weighing under 1500 g by 85% (OR: 0.15; 95% CI: 0.08–0.26; p < 0.0001) [5–13,27]. Candida-related mortality was decreased by 96% (OR: 0.04; 95% CI: 0.01–0.31; p = 0.0055) and overall mortality rate was reduced by 25% (11% in the fluconazole-treated infants compared with 16.3% in the control patients; OR: 0.75; 95% CI: 0.58–0.97; p = 0.029). Healy et al. also reported the elimination of Candida-related mortality in any patient in their NICU when fluconazole prophylaxis was targeted to infants weighing under 1000 g [8,14].

Fluconazole prophylaxis is extremely cost effective. Uko et al. examined the cost of fluconazole prophylaxis and showed a significant cost benefit of US$516,702 over 18 months in their NICU [12]. At our institution, the pharmacy cost of one dose is US$18 (Marcia Buck, PharmD, Pers. Comm.), making the cost of the average time of prophylaxis of 4–6 weeks (8–12 doses) between US$144 and US$216 per patient.

Resistance & safety

Some of the issues related to prophylaxis include side effects and resistance. With any antimicrobial therapy, resistance needs to be examined. There are important differences between the development of resistance in bacteria and fungi. Most importantly, fungi do not have mechanisms of transferring resistance between fungi and lower dosing is not associated with emergence of resistance [28,29]. In bone marrow transplant patients, fluconazole prophylaxis has been successful for almost 20 years in decreasing both ICIs and mortality, while azole-resistant isolates have remained low, at approximately 5% [30,31]. Resistance has been observed in adult studies (HIV-positive patients) with prolonged courses (3–24 months) of fluconazole and when escalating to higher doses (400–800 mg) [28]. By contrast, adult studies employing lower doses (150–200 mg) and weekly doses have been successful in preventing fungal infections without the development of resistance [32,33].

Neonatal prophylaxis studies have not reported a significant change in or emergence of resistant species over the course of prophylaxis during the study period of 2–3 years or over a 5-year period encompassing two studies [5,7]. Furthermore, there was no emergence or increase in the incidence of colonization or infection due to Candida glabrata or Candida krusei reported in any study as well as a recent single-center analysis of 10 years (4 years prior to and 6 years following fluconazole prophylaxis) [34].

Fluconazole prophylaxis at higher doses (6 mg/kg or greater) and frequency may be associated with the development of some resistance [35,36]. Two studies in preterm infants and one in preterm baboons have reported this association [35–37]. Sarvikivi et al. have reported the use of fluconazole prophylaxis over a 12-year period, finding no C. glabata or C. krusei infections and only two cases of resistant C. parapsilosis BSIs. In their NICU, fluconazole prophylaxis was most commonly administered daily at a dosage of 6–12 mg/kg to infants weighing under 1000 g and aged less than 30 weeks gestation between 1991 and 2000. No fungal resistance was detected during that time period; however, when the use of fluconazole was broadened to the entire NICU in 2000, the emergence of resistance was noted and associated with two BSIs [35]. Both infected patients were infected with the endemic strain that had previously been susceptible. One important distinction in this series was that fluconazole was used for both prophylaxis and treatment of fungal infections in this NICU. Furthermore, higher doses of daily fluconazole are associated with development of resistance and may have been a factor in this study [28]. These observations suggest that lower and less frequent dosing (3 mg/kg twice a week) may delay or attenuate the development of resistance and still prevent serious infections [7].

Parikh et al. recently reported in a single-center study that fluconazole prophylaxis (6 mg/kg every 72 h until day 7, then daily until day 28) decreased colonization but not ICI, with non-albicans species responsible for 96.8% of infections and the majority due to C. glabrata. Prior to the prophylaxis study period, fluconazole was used for treatment in their NICU with rates of ICI of approximately 25% in infants weighing under 1500 g, meaning that there was a high exposure rate of treatment doses per patient and possibly this total dose exposure in their NICU correlated with the selection of C. glabrata or other species with decreased susceptibility to fluconazole. If a NICU has been primarily using fluconazole for treatment or empiric therapy and has high rates of infection, fluconazole prophylaxis may not be effective. Possibly changing primary treatment and empiric therapy to one of the amphotericin B formulations for 6–12 months before instituting prophylaxis may change the NICU Candida flora, since resistance to azoles decreases when fungi are not continuously exposed to them [29]. Additionally, amphotericin B may also eradicate some of the resistant species from the horizontally transmitted flora. Alternatively, in those NICUs, nystatin prophylaxis may be a better choice and its efficacy should be further studied.

These findings suggest that it is important to focus the use of fluconazole prophylaxis to only selected high-risk NICU patients and primarily for prophylaxis, with a different antifungal chosen for treatment or empiric therapy, such as amphotericin B when able, as this will limit overall fungal exposure to fluconazole and possibly prevent the emergence of resistance [26].

Side effects of fluconazole prophylaxis

In the RCTs, there were no significant differences in bacterial infections, necrotizing enterocolitis or liver function. Safety regarding long-term neurodevelopmental outcome is also favorable [38]. One of the retrospective studies reported a higher incidence of cholestasis in the fluconazole prophylaxis patients, which was transient with no difference at discharge, while another retrospective study demonstrated a lower incidence of cholestasis [12,13]. Since there were no significant differences in direct bilirubin or liver enzyme levels in the four randomized placebo-controlled trials, it may be that other factors present during the study period increased the likelihood of cholestasis [14].

Nystatin

Nonabsorbable nystatin prophylaxis aims to prevent or reduce GI tract fungal colonization. The first RCT of nystatin prophylaxis (100,000 units orally, three-times daily) was studied in ventilated preterm infants weighing under 1250 g until 1 week after they were extubated and demonstrated a decrease in Candida UTI but no effect on the incidence of fungal BSI on subanalysis [39]. Candida BSI occurred in none of the 33 nystatin-treated patients compared with two of the 34 (6%) placebo-treated patients (p = 0.25). Fungal UTI occurred in two (6%) nystatin and ten (29%) control patients (p = 0.01).

A recent prospective single-center study examined nystatin prophylaxis (100,000 units three-times daily orally or via a nasogastric tube) in both preterm and term NICU patients, excluding infants with congenital defects requiring surgery [40]. Examining the study infants by birth weight, 24% (n = 948) weighed under 1500 g, 8.7% (n = 349) weighed under 1000 g and 0.8% (n = 30) weighed under 750 g. Infants were randomized into two groups: group A had oral fungal cultures performed and received nystatin prophylaxis only if oral Candida colonization was isolated and group B infants were given prophylaxis regardless of colonization status. In group A, infants weighing under 1500 g who did not receive nystatin prophylaxis had a 36% probability of developing of candidemia (131 out of 358), compared with an incidence of 13.9% (16 out of 115) in group A infants given prophylaxis only if oral colonization was detected (p ≤ 0.001). In group B, in which nystatin prophylaxis was started in the first 72 h of life regardless of the colonization status, only 3.6% (17 out of 475) developed invasive infection (p ≤ 0.001, compared with group A who received nystatin prophylaxis). Nystatin was started at 12.6 ± 2.4 (mean ± standard deviation [sd]) days in the subgroup of group A receiving prophylaxis and at 2 ± 0.5 days in group B. Prophylaxis did not decrease infection in infants who had a birth weight of over 1500 g. Nystatin prophylaxis initiated in the first 72 h of life had lower rates of infection compared with starting nystatin prophylaxis after oral colonization was detected (3.6 vs 13.9%; p = 0.01), demonstrating the added benefit of using prophylaxis early after birth and, thus, before most colonization has occurred. This decreased efficacy of nystatin prophylaxis when initiated after colonization has also been reported in other studies [41–44]. Limitations of the study included the lack of a placebo group, the fact that there were few infants weighing under 750 g and that infants with congenital defects requiring surgery (e.g., gastroschisis) were excluded. These are high-risk groups in which efficacy must be studied.

Future research

Although the efficacy of nystatin prophylaxis is less clear, a four-arm study examining fluconazole, nystatin, fluconazole plus nystatin, and placebo would answer questions of efficacy and safety in extremely preterm infants. Fluconazole plus nystatin may aid in the prevention of infection by the few isolates with decreased azole susceptibilities. The approach of administering fluconazole prophylaxis only during antibiotic treatment may be effective but needs further study in a RCT. Prophylaxis started after the infant is colonized may not be as effective as starting prophylaxis shortly after birth. Infants weighing over 1000 g with gastrointestinal disease also have a high rate of developing ICI and a RCT in this population is also needed [40].

Who should receive antifungal prophylaxis?

The question many have raised is ‘who would benefit from receiving antifungal prophylaxis?’ [45] Approximately 30–34% of NICUs have instituted antifungal prophylaxis in the USA and UK [21,45]. Several factors should influence that decision, including incidence, mortality and NDI.

• • Targeted prophylaxis should be given to all infants weighing under 1000 g and/or aged 27 weeks or under while they require intravenous access (peripheral or central) starting at birth and continuing for up to 6 weeks of life. This subpopulation of preterm infants has high mortality and NDI, and this approach has demonstrated efficacy and safety with fluconazole without the emergence of resistance in randomized placebo-controlled trials, while eliminating Candida-related mortality;

• • Even in a NICU with low overall rates of ICI (<2%), infants aged 27 weeks or under are likely to be at a high risk and would benefit from prophylaxis. Incidence and outcomes by gestational age should be examined and tracked (Figure 1). ICI can be analyzed by filling out the chart in Figure 2 at institutions with low rates to determine the gestational age range in which ICI occurs and to identify those infants who should receive prophylaxis. There is likely a gestational cut-off age wherein ICI falls to zero. If NICUs do not have neurodevelopmental outcome data, prophylactic treatment of high-risk infants weighing under 1000 g or aged 27 weeks or under should be strongly be considered, since treatment of documented infections does not always prevent the NDI and mortality of these infections [3,4];

• • NICUs with high rates in infants weighing 1000–1500 g, may choose prophylaxis in these patients. A targeted approach to infants with a central venous catheter or on antibiotics for more than 3 days has been used in retrospective studies [9,12].

Fluconazole or nystatin?

From the aforementioned data, there is overwhelming evidence in support of fluconazole prophylaxis and the data are consistent with the recent Cochrane review [46]. The Cochrane review data revealed a statistically significant reduction in invasive fungal infections in infants who received prophylaxis (typical relative risk: 0.23; 95% CI: 0.11–0.46), with a number needed to treat of nine (95% CI: 6–17) [46]. Further study of nystatin prophylaxis is needed in extremely preterm infants as, thus far, there is only one RCT in infants weighing under 1250 g [39].

Epidemiologic studies including critical information regarding antifungal prophylaxis and data by each gestational age and 100 g birthweight group are needed. Without this information, these studies may have minimized the incidence of ICI and missed important risk factors if infants were receiving antifungal prophylaxis. Nystatin prophylaxis in preterm infants is common and may explain the lower reported rates in the UK and Australia and the varying incidence among NICUs in the USA [21,45].

Dosage & schedule

Fluconazole administered at 3 mg/kg doses intravenously twice weekly until intravenous access (peripheral or central) is no longer needed has efficacy and the best safety profile. Manzoni et al., in their multicenter RCT, demonstrated that 3 or 6 mg/kg doses are equally effective [47]. However, dosing with 3 mg/kg is preferable for two reasons: first, drug concentrations in the skin, lung and mucous membranes are greater than plasma levels (therefore, larger doses may be unnecessary); and second, the use of higher doses may foster the development of fungal resistance. Furthermore, the goal of prophylaxis is to use the lowest effective dose (usually 50% of the treatment dose). At our institution, we administer fluconazole prophylaxis twice weekly on the same days – every Tuesday and Friday – at a specified time (e.g., 10:00), which further reduces pharmacy costs and limits medication errors. Therefore, 3 mg/kg given twice weekly is the optimal dosing schedule to maximize efficacy, safety and cost.

The time is now

Pediatrics has led the way in infectious disease prevention and flucanazole prophylaxis can eliminate one cause of nosocomial infection in preterm infants and it should be instituted in every NICU. With single-center and multicenter RCTs and a meta-analysis demonstrating a 91% decrease in ICI in infants weighing under 1000 g and a 96% decrease in mortality, fluconazole prophylaxis should be targeted to this group of infants weighing under 1000 g or aged 27 weeks or under, due to the high rate of mortality and NDI. The prevention of ICIs in extremely preterm infants also eliminates Candida as a cause of mortality and NDI in these vulnerable hosts.

Pediatric infectious diseases and neonatal organizations can help guide NICUs and support good epidemiologic follow-up of ICI, fungal susceptibilities and mortality. The 2006 Report of the Committee on Infectious Diseases of the American Academy of Pediatrics (Red Book^®) has a statement supporting the use of fluconazole prophylaxis in high-risk preterm infants under 1000 g [48].

In striving for better outcomes for our most extremely preterm infants, from prenatal care to the delivery room to discharge, infection prevention is critical to decreasing many of the associated morbidities and the mortality rate of these infants. At this time, the benefits of antifungal prophylaxis significantly outweigh any risks. Every day we wait, we place our smallest and most vulnerable infants at risk of harm from ICI: leading to either death or long-term NDI. We must protect our infants now.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.